What is an ontology, in business terms?

An agreed definition of the things your business deals with, what they are called, how they relate to each other and which rules apply to them, held in a form both people and systems can read. For a wealth firm that means clients, households, products, wrappers, providers, advisers, services, fees, reviews and the relationships between them, plus the rules that govern each. It is a dictionary and a rulebook combined, maintained once rather than rediscovered every time a new system or a new colleague arrives.

What is a knowledge graph, and why does it matter for AI?

A knowledge graph is the ontology populated with your actual data: not just "a client has holdings with providers" but this client, these holdings, these providers, with every link traceable. It matters for AI because it is the difference between a model that answers from general knowledge and one that answers from your firm's knowledge, with the protected tax-free cash flag attached to the client it belongs to. It is also what lets you show why an answer was given, which the FCA's expectations on explainability and accountability increasingly require. [3]

What are AI guard rails?

Limits and rules applied to what an AI system can do, say and act on, drawn from your organisational and regulatory context. Some are hard stops: never state a tax-free cash figure without checking for protection; never suggest a consolidation under targeted support. Some are routing: escalate to a human when the confidence is low or the customer is flagged as vulnerable. Guard rails are only as good as the context they are built from. A guard rail that does not know protection exists cannot protect anyone.

How do you encode business rules for automation?

Start by finding them, which is harder than it sounds, because most live in people and documents. Write each one down in plain language with an owner, a source and an example. Resolve the conflicts, of which there will be more than you expect. Then and only then put them into a form a system can apply, and keep the plain-language version as the master so that a human can always check what the machine was told. The order matters. Firms that start with the technology end up encoding whatever the system vendor assumed.

Isn't this just documentation?

Documentation describes the rules for a reader. Context, in the sense used here, defines them in a form that is checked and applied, by people and systems alike, so that the document and the behaviour cannot drift apart without someone noticing. The 2011 product guide in the story above was documentation. Nobody read it, and the system could not.

Are organisations struggling to demonstrate value from their AI technology investments?

11 minutes ago

6 min read

Boards and leadership teams in advice and wealth firms are facing more questions from the regulator on how they are using AI technologies. This October, the FCA said AI adoption in financial services was already here [1], and agentic AI is the next investment many firms are now looking to implement.

Organisations are showing they are open to agentic AI, but they draw a clear line in how far they trust it. In a survey of 178 advisers published in August by GBST and the Lang Cat, 62% were comfortable with agentic AI being built into the platform they already use. Comfort varied by task, with 77% happy for it to handle onboarding, while only 43% felt the same about pension transfers, where client money is directly at stake [2].

The opportunity and the risk of AI technology investments

Agentic AI is technology that can carry out work autonomously, completing a sequence of tasks and making decisions within the rules and guardrails that have been set, while employees can review and approve the outcome. Many tools in use today respond to a single request, such as drafting a note or summarising a document, and then wait for a person to decide what happens next. An agentic workflow operates through the steps of a process, deciding what to do next until the task is complete or something needs human attention. It is the next generation of automation.

For advice and wealth firms, that means much of the administration around the advice can automated and help improve client outcomes and experience. The tools to build and run agents are now accessible to firms of every size, which opens the opportunity to far more of the market than a few years ago.

The risk is where the implementation is not built on the right foundations and being applied to the wrong use cases. Any work executed by an agent on a client’s portfolio carries the firm’s name, and under SM&CR a named individual remains accountable for it. That person needs to know what the agent is permitted to do, how any outcomes were determined and how its work is being assured.

Trust in the system outputs come from how the agents are implemented, ,tested and the firm organises the work around it, which comes down to the operating model It is not uncommon for projects to implement without considering the broader operating model, this is the focus of Taking Advantage of Agentic AI, the second session of the Big Shift Season 2, on Wednesday 18 November.

Alirity webinar poster with two speaker headshots, promoting The Big Shift Season 2 on Wednesday 18th November, 12:30–13:30 GMT.

Building on the Single Customer View session

Our early October session looked at Addressing the Single Customer View Challenge, where Stuart Coleman from Tetmon joined Poppy Achilles and Charlie Symonds to look at how you can bring customer data into a single view while keeping control of it. The conversation covered what data sovereignty means for regulated firms and explored a decentralised approach, which queries data where it already sits to give a joined-up view without moving everything into one central platform.

Two points from that session connect directly into this one. As AI tools become easier for people across a firm to acquire, they can reach customer data in ways the firm has not planned for, and the records they bring together still need a person to review them. An agent working on a client's case depends on the same foundations: authentic data the firm can trust and control, and clear limits on what the technology can do with it.

Agentic AI in a SIPP new business flow

Alirity has built a working agentic prototype for a self-invested personal pension (SIPP) new business flow.

A client decides to bring several pensions together, and before any advice can be given, the case moves through a long run of administration: collecting documents, running identity checks, completing the application, entering the same details into several systems and chasing each existing provider to release the funds. In some cases, transfers can take weeks, and throughout that time someone in the team is following up with providers and keeping the client informed.

Automation can take on much of this administration, keeping the case moving and bringing the adviser in whenever something needs their judgement. The client hears about progress without having to ask, freeing up time for higher value tasks. Over time, the hours saved compound with every case, increasing organisational capacity to support growth and likely improving customer outcomes and experience.

Some steps call for judgement, such as details that do not match between providers or a client showing signs of vulnerability, as does any decision that affects the client's money, including whether a transfer goes ahead. With a human in the loop, these stay with the adviser, who reviews and approves the work before the case moves on, and the process remains led by the people who know the client. In the session, Charlie will walk through the prototype to show how this works in practice for you to apply to your own AI technology investments.

Hand using tablet with glowing AI Agent workflow diagram, code, lightbulb, gears, cloud and check icons on dark blue background.

How do we ensure we are achieving value?

Like any change, it is important to understand the value and benefits it will bring. This comes through understanding the key value drivers and measures. We would see that sitting across:

Improving the cost to serve
Improving the experience and engagement with customers
Reduction in quality issues, rework and complaints
Supports sustainable growth
Reduction in risk exposure

This can only be done by having the right foundations in place to build & run the implementation alongside the measures to demonstrate return on investment against your agreed value measures.

The value is only likely to be achieved if the changes are done holistically and the operating model shifts as well.

Join us on the 18th of November

The second session of the Big Shift Season 2 brings this together. We will explain the foundations agentic AI needs to work, walk through an example SIPP new business flow and close on how you can get started alongside potential use cases. Poppy Achilles will host the session and open the floor to questions.

The session is designed for leaders and practitioners in financial organisation, from executives and transformation leads to heads of operations, data and technology. It is free to attend for anyone working in the sector.

Taking Advantage of Agentic AI
Wednesday 18 November 2026, 12:30 to 13:30 GMT

Questions people are asking

What is agentic AI?

Agentic AI is technology that carries out work itself, completing a sequence of tasks and making decisions within the guardrails a firm sets.

How is agentic AI different from other automation?

Rules-based automation follows predefined steps and stops when something falls outside them. Agentic AI works towards a defined goal, choosing its next step based on the situation in front of it and using the tools and data a firm has approved. It still works within the limits set, and people remain accountable for the outcome.

Why do AI projects stall?

Many stall for organisational reasons. When an agent is added to an existing process without changes to how the work is organised, cases still pass through the same checks and the same people, and the time it saves does not reach the team. Rules that have never been written down and client data spread across systems make it harder still.

What does an operating model for agentic AI involve?

It covers the decisions that allow an agent to work safely, such as which steps it handles, where people review and approve its work, and who is accountable for the outcome. This sits alongside the primary capabilities that support the model and the relevant changes to the organisation design and capability. Our Organisation Rewired programme works with firms to put these foundations in place.

How does agentic AI fit with SM&CR and Consumer Duty?

Accountability stays with the firm and its people. Under the Senior Managers and Certification Regime, a named individual remains responsible for each area of the business, including work an agent carries out, and Consumer Duty asks firms to evidence good client outcomes. Clear limits on what the agent can do and a record of each step it takes help firms meet both.

What is the Big Shift?

The Big Shift is a webinar series run by Alirity for the financial sector, with each season focusing on key challenges facing the industry. Season 2 covers Single Customer View, explored in October, and agentic AI, on 18 November.

When is the session, and who is it for?

Taking Advantage of Agentic AI takes place online on Wednesday 18 November 2026, 12:30 to 13:30 GMT. It is designed for leaders and practitioners in financial services, including executives, transformation and change leads, heads of data, technology and operations, and advisers.

How do I sign up?

Register through the sign-up form to receive the joining link. The session is free to attend for anyone working in the advice and wealth management sector.